An application of Z-Box method in dairy cow feedingto estimate the relationships among peNDF, otherfeed variables and productive data

Physically effective NDF (peNDF) is defined as the fraction of fibrethat stimulates chewing and contributes to the floating mat oflarge particles in the rumen, and consequently to its regular activity.PeNDF is calculated from a physical effectiveness factor (pef),varying from 0 (NDF stimulates no chewing) to 1 (max chewing),which may be obtained by laboratory-based particle sizing techniques,such as Penn State Particle Separator, Mertens Separator,Z-Box, Cut Accuracy Test, based on the proportion of DM retainedon sieves (by horizontal or vertical shaking). We chose Z-Boxmethod, thanks to its easy use and applicability to as-is feed andtotal mixed rations (TMR), and we are trying to obtain an estimatingequation which may predict milk fat content and/or other productivedata from peNDF and other variables measured on TMR.To this aim, samples of TMR collected from several farms aresieved (3 sub samples each), and undergo proximate analysis,NDF, ADF, ADL and starch. Milk yield, milk fat, water addiction toTMR are collected on farm; qualitative data such as type of forage,breed, season, geographical origin and altitude (plain/hill/mountain)are also taken into account, to estimate their possible effect.As a first step, in order to investigate the complex relationshipsexisting among this wide set of variables, Principal ComponentAnalysis (PCA) is used as a data exploration tool. Two PCA models(presence of silage or not in TMR) are calculated separately. Foreach PCA model, the overall correlations among all the consideredvariables and their relative importance are investigated by meansof the loadings plots, posing particular attention to the correlationswith peNDF and with milk fat. Moreover, it is also possible to identifyhow the different groups of samples depend on specific variables.